Unexpected arousal, anxiety sensitivity, and their interaction on CO₂-induced panic: Further evidence for the context-sensitivity vulnerability model

Abstract

The present experiment tested several predictions derived from the context-sensitivity vulnerability model of panic. Participants (N = 79) scoring either high or low in anxiety sensitivity (AS) and with no history of unexpected panic were randomly assigned to one of two instructional sets: expected arousal (EA) or expected relaxation (ER). All participants were administered inhalation of room air and 35% CO₂ in a counterbalanced order. Consistent with theoretical predictions, High-AS participants who received ER instructions showed greater emotional responding compared to High-AS participants who received EA instructions, while instructional set did not affect responding among Low-AS participants. Panic attacks were observed in 52% of the High-AS-ER group compared to 17%, 5%, and 5% in the High-AS-EA, Low-AS-ER, and Low-AS-EA groups respectively. These findings are consistent with the theory's assertion that dispositional tendencies, such as anxiety sensitivity potentiate the panicogenic effects of threat-relevant context variables.

Introduction

Laboratory provocation of panic attacks has been widely used as a means for investigating the pathogenesis of panic disorder (Margraf et al., 1986, McNally, 1999). In this paradigm, a panic disorder group and a normal or psychiatric control group are administered an agent that induces somatic perturbations. Demonstration of greater challenge-induced panic among the panic disorder group has been frequently cited as evidence implicating neurobiological dysregulation in the pathogenicity of panic disorder. Numerous challenge agents have been shown to induce panic attacks in patients with panic disorder but rarely in normal controls. These include sodium lactate (Cowley and Arana, 1990, Liebowitz et al., 1984), yohimbine (Charney, Heninger, & Breier, 1984), carbon dioxide (Griez, deLoof, Pols, Zandbergen, & Lousberg, 1990), caffeine (Charney, Heninger, & Jatlow, 1985), cholecystokinin tetrapeptide (Bradwejn, Koszycki & Shriqui, 1991), and hyperventilation (Holt & Andrews, 1989).

In contrast to biological explanations (Gorman et al., 2000, Klein, 1993), several prominent psychological formulations of panic disorder have emerged over the past decade (Barlow, 1988, Beck and Emery, 1985, Bouton et al., 2001Clark, 1986, Goldstein and Chambless, 1978, McNally, 1990, Wolpe and Rowan, 1988). These converge in positing a core psychopathological feature, namely the tendency to respond fearfully to benign somatic cues. The theories diverge, however, in the presumed mechanisms accounting for this tendency i.e., interoceptive conditioning (Barlow, 1988, Goldstein and Chambless, 1978), catastrophic misinterpretation (Beck and Emery, 1985, Clark, 1986); or an enduring dispositional variable such as anxiety sensitivity (McNally, 1990, McNally, 2002, Reiss, 1991).

Support for these psychological models comes from studies showing a linkage between the predisposition to perceive anxiety as harmful (i.e., anxiety sensitivity) and panic disorder. Panic disorder patients show elevations on measures tapping fear of fear, such as the Agoraphobic Cognitions Questionnaire (Chambless, Caputo, Bright & Gallagher, 1984) or the Anxiety Sensitivity Index (ASI) (McNally, 2002, McNally and Lorenz, 1987, Reiss et al., 1986, Telch et al., 2003). Anxiety sensitivity predicts the diagnostic severity of panic disorder (Jones & Barlow, 1991) and behavioral fear responding to voluntary hyperventilation predicts agoraphobia status among panic disorder patients (Telch et al., 2003). Moreover, elevated anxiety sensitivity normalizes after successful cognitive-behavioral treatment for panic (Smits et al., 2008, Smits et al., 2004, Telch et al., 1993), as does emotional responding to inhalation of 35% CO₂ gas (Gorman et al., 2004, Schmidt et al., 1997a).

The aforementioned studies do not rule out the possibility that elevated anxiety sensitivity is a concomitant or consequence of panic disorder. A more stringent test of whether anxiety sensitivity operates as a risk factor in the development of panic is to demonstrate a greater panic proneness among participants displaying elevated anxiety sensitivity but who have not yet developed panic disorder. Several prospective studies have demonstrated that people who score high on the ASI are at greater risk for developing naturally occurring panic attacks and related anxiety disorders compared to those who score low on the ASI (Maller and Reiss, 1992, Schmidt et al., 1997a, Schmidt et al., 1997b, Schmidt et al., 1999, Schmidt et al., 2006a, Schmidt et al., 2006b). Similarly, causal modeling studies of learning history suggest that some early learning experiences may influence development of anxiety sensitivity which in turn results in a higher risk of panic attacks (Stewart et al., 2001).

Administration of provocation agents that reliably induce intense somatic reactions also provides a useful research paradigm for investigating the interplay between dispositional and contextual factors in panic. Telch and colleagues (Telch, 1995, Telch et al., 2010) have proposed a model for fear responding in laboratory studies of panic provocation and naturally occurring false alarms. More specifically, their context-sensitivity vulnerability model posits that dispositional factors such as anxiety sensitivity potentiate fear in response to a threat-relevant context. Here, context is defined as any stimulus that influences one's perception of threat and can therefore be internal such as a somatic cue (e.g., chest tightness), a thought (I’m going to lose control), or emotion (e.g., anger); or external (e.g., being in a densely crowded place with no exit nearby). Potentiating dispositional sensitivities can be quite broad as in the case of trait anxiety or more narrow as in the case of anxiety sensitivity or even more narrow in the case of cardiac or respiratory sensitivity. The model further posits that degree of fear responding depends on the conceptual match between the dispositional sensitivity profile of the person and the threat-relevant context. This latter prediction has been termed the context-sensitivity matching hypothesis (Telch et al., 2010).

Experimental manipulations of context during panic provocation have been investigated in several studies (cf. Zvolensky & Eifert, 2001). Initial studies provided evidence for the main effects of context. For example, using an illusory control paradigm during a 35% CO₂ challenge, Sanderson, Rapee and Barlow (1989) demonstrated that compared to PD patients who received no illusion of control, those who were led to believe that they could control the concentration of CO₂ gas displayed a significantly lower probability of CO₂-induced panic. Similarly, Rapee, Mattick and Murrell (1986) found that in contrast to PD patients who were provided a full explanation of the effects of a 50% CO₂ inhalation, PD patients who were provided no such explanation displayed a greater proportion of catastrophic cognitions, and a higher likelihood of panic.

Several studies have since provided evidence for the interaction between dispositional tendencies and contextual factors on fear responding to biological challenges. For example, Schmidt and Trakowski (1999) manipulated attentional focus by instructing participants to focus on either internal cues or external cues during a single inhalation of 35% CO₂ gas. While there was no significant interaction between patient status and attentional focus, post hoc analyses revealed that attentional focus was a significantly stronger predictor of fearful responding among participants high in anxiety sensitivity. A more stringent test of the hypothesis that contextual and dispositional factors play a causal role in the psychopathogencity of panic comes from studies of individuals who have not yet experienced panic disorder or unexpected panic attacks. In a caffeine challenge experiment with nonclinical subjects scoring high or low on anxiety sensitivity, Telch, Silverman, and Schmidt (1996) manipulated a threat-relevant contextual factor (i.e., availability of an ostensible caffeine antidote) following participants’ ingestion of 450 mg of caffeine. Consistent with prediction, the context manipulation (availability of the antidote) strongly influenced fear responding to the caffeine challenge for subjects high in anxiety sensitivity but not for those low in anxiety sensitivity. Similar findings were reported by Zvolensky and colleagues in their experimental manipulation of offset control during 20% CO₂ inhalation (Zvolensky et al., 1999, Zvolensky et al., 1998, Zvolensky et al., 2001).

Initial evidence for the context-specificity matching hypothesis was provided by another recent experiment from our group (Telch et al., 2010). In this study, we manipulated presence or absence of a cardiac defibrillator during 35% CO₂ inhalation. Consistent with prediction, increased fear responding in the defibrillator condition was observed only among nonclinical participants reporting high cardiac sensitivity (Telch et al., 2010). Indeed, interactions between context (defibrillator present [yes, no]) and other dispositional tendencies (e.g., respiratory sensitivity, trait anxiety, depression, anxiety sensitivity) were not significant.

In the present experiment, participants scoring either high or low in anxiety sensitivity with no history of panic disorder or unexpected panic attacks were administered single inhalations of 35% CO₂-enriched air and regular room air in a counter-balanced order. In addition, to manipulating the content of the gas mixture (35% CO₂-enriched air vs. room air), we also manipulated participants’ expectations concerning the effects of the gas inhalation by providing challenge instructions that created the expectation that the gas mixture would be either physically arousing (expected arousal), or relaxing (unexpected arousal). Consistent with the potentiation hypothesis from the context-sensitivity vulnerability model, we hypothesized that (a) anxiety sensitivity would potentiate the effects of the two context manipulations (content of gas mixture and expectedness of arousal) on fear responding. Consistent with the threat appraisal hypothesis from the model we expected that participants who perceived their CO₂-induced arousal as unexpected (i.e., threat enhancing context) would display greater subjective fear and a higher probability of panic compared to those who perceived their CO₂-induced arousal as expected (i.e., threat attenuation context). Finally, consistent with the theory's context-specificity matching hypothesis, we predicted that anxiety sensitivity – because of its more specific conceptual match to the threat-relevant context – would potentiate the effects of the arousal expectedness manipulation on fear responding to CO₂ inhalation. No such potentiation effect was expected for less specific dispositions such as trait anxiety or depression.